At present, with the progressive development of somatosensory recognition technology, somatosensory recognition has been applied in people's daily life more and more. Somatosensory recognition is mainly used to recognize and obtain a body's positions in movement, such as head, arms, or feet, and then implement man-machine control by means of auxiliary equipments depending on the data of the obtained positions. The somatosensory recognition may be applied in environment, such as in playing games, in control of TV set, and in the office. In case of TV somatosensory games as an example, people can play games easily while watching TV, and realize better man-machine interaction effect with the TV as a bearer.
However, in main stream somatosensory recognition systems, most of them capture body's action with image pick-up elements, and participators need to stand in just front of the image pick-up elements for capture an image accurately. However, there is certain error with the above-mentioned method. For example, while the body moves in three directions, i.e., Z (such as up-down direction relative to the image pick-up element), Y (such as left-right direction) and X (such as front-rear direction) directions, since the image pick-up element is not sensitive to the movement in Z direction, misjudgments tend to occur in the Z direction parallel to the image pick-up element. In particular in case of an arm as the recognized object, if the arm is blocked by the body, the conventional somatosensory recognition system can not acquire the arm's position, thereby failing to determine body's action accurately.